The Job Thief: A Dragon's Tale

Summary

This chapter tells the story of a well-meaning dragon named Algorithm who keeps accidentally automating everyone's jobs while trying to be helpful — a sympathetic villain origin story. Through this narrative, students explore the mechanics of tech hype (breakthrough announcements, AI washing, investor pitch decks), the startup mythology that fuels disruption narratives, and the beginning of job displacement and automation anxiety. The dragon would like you to know it's not personal.

Concepts Covered

This chapter covers the following 10 concepts from the learning graph:

AI Demo vs Product
Breakthrough Announcement
AI Washing
Moving the Goalposts
Investor Pitch Deck
Startup Mythology
Disruption Narrative
Job Displacement
Automation Anxiety
Workforce Disruption

Prerequisites

This chapter builds on concepts from:

Welcome, Colleagues

Let me be perfectly clear. This chapter introduces a dragon who automates jobs. The dragon is not evil. The dragon is efficient. The distinction, as you will discover, provides remarkably little comfort to the villagers.

The Dragon Arrives

Once upon a time — which in technology means approximately eighteen months ago — a dragon arrived at the village of Industria. The dragon's name was Algorithm. Algorithm was not the kind of dragon that hoarded gold in mountain caves. Algorithm was the kind of dragon that looked at your workflow, identified inefficiencies, and eliminated them before you finished your coffee. Algorithm meant well. This was the problem.

Algorithm had been raised in a laboratory by researchers who believed that making processes faster and cheaper was inherently good. They had not asked faster and cheaper for whom. They had not considered what happens to the people whose processes are made unnecessary. They had published a paper titled "Toward Optimal Efficiency in Document Processing" and attached a demo video that showed Algorithm summarizing legal contracts in 0.3 seconds. The paper received 4,000 citations. The legal assistants received layoff notices.

This is the story of how disruption works. It is not a story about malice. It is a story about incentives, about the gap between a demo and a product, and about what happens when the people building the dragon are not the people living in the village.

AI Demo vs Product: The Gap That Eats Careers

The AI demo is the technology industry's most powerful sales tool. It is a carefully controlled demonstration of what an AI system can do under ideal conditions — with curated inputs, pre-selected examples, and a presenter who knows exactly which questions to ask and which to avoid.

The AI product is what the customer actually receives. It is the system deployed in the real world, processing messy data, encountering edge cases, and performing in conditions the demo never contemplated. The gap between the demo and the product is the space where careers, budgets, and business plans go to die.

Key differences between demos and products:

Feature	AI Demo	AI Product
Input data	Hand-selected, clean	Messy, incomplete, contradictory
Edge cases	Carefully avoided	Encountered constantly
Performance	Exceptional (cherry-picked examples)	Variable (average over all cases)
Error handling	Errors edited out of video	Errors presented to customers
Cost	Free (it's marketing)	Expensive (it's infrastructure)
Audience	Investors and journalists	Users with actual problems

Algorithm the dragon had a spectacular demo. In the demo, Algorithm could read a 200-page contract and produce a summary in seconds. In the demo, Algorithm never misread a clause, never hallucinated a provision, and never accidentally summarized the wrong document. In the demo, Algorithm was perfect.

In production, Algorithm misread ambiguous language 12% of the time, occasionally summarized the terms of service from a different company entirely, and once produced a contract summary that included a recipe for lemon chicken — a hallucination that, in fairness, was more useful than the original contract.

The gap between demo and product is not a bug. It is the business model. The demo sells the investment. The product is what gets built after the money arrives. In many cases, the product never fully closes the gap — but by then, the demo has served its purpose, and the investors have moved on to the next dragon.

Breakthrough Announcements: The Language of Hype

A breakthrough announcement is a press release, blog post, or media event in which a company or research group declares that it has achieved something unprecedented in artificial intelligence. The announcement typically includes phrases such as:

"State-of-the-art performance"
"Surpasses human-level accuracy"
"A major step toward AGI"
"This changes everything"
"We are humbled by these results"

Breakthrough announcements follow a formula that has remained remarkably stable since the field's inception:

Overstate the achievement: "Our model achieves 97% accuracy" (on a benchmark specifically designed to make the model look good, using a metric that does not reflect real-world performance)
Understate the limitations: The announcement will mention limitations, if at all, in the final paragraph, using language designed to be skipped ("While further work is needed in certain edge cases...")
Imply transformative impact: The announcement will connect the specific technical achievement to a grand vision ("This brings us closer to systems that can understand and reason about the world")
Omit what was tried and failed: Breakthrough announcements are survivor narratives. The dozens of approaches that did not work are not mentioned. The impression is one of inevitable progress, not of trial and error

The frequency of breakthrough announcements in AI has increased from approximately one per month in 2020 to several per week in 2025. Either the pace of genuine breakthroughs has accelerated by a factor of 20, or the definition of "breakthrough" has been diluted to the point of meaninglessness. The data supports the latter interpretation.

A Critical Observation

The data is unambiguous. If breakthrough announcements in AI were accurate, the field would have achieved approximately 347 paradigm shifts since 2020. The actual number of paradigm shifts is between one and three, depending on how generously one defines the term.

AI Washing: The New Greenwashing

AI washing is the practice of labeling products, services, or companies as "AI-powered" when the AI component is minimal, non-functional, or entirely absent. It is the direct descendant of "greenwashing" — the practice of marketing products as environmentally friendly without meaningful environmental benefit — updated for the current hype cycle.

Common forms of AI washing include:

The API call: A product that sends user data to a third-party AI service and displays the result is marketed as "our proprietary AI." The company's contribution is a button that says "Analyze"
The rebrand: A product that uses basic statistical methods (regression, averaging, sorting) is rebranded as "AI-driven analytics." The mathematics has not changed. The marketing has
The future tense: A company announces "AI capabilities" that are "coming soon" or "in development." The AI exists in the same way that the unicorn exists — in the pitch deck
The sprinkle: A product adds a chatbot to its help page and describes itself as "AI-first." The chatbot answers three types of questions. All other questions receive: "I'm sorry, I didn't understand that. Please contact support"
The benchmark: A company reports performance on a specific benchmark that was selected because the product performs well on it. Other benchmarks, on which performance is mediocre, are not mentioned

AI washing works because most people — including most investors, most journalists, and most executives — cannot evaluate whether a product's AI claims are substantive. The label "AI-powered" functions like the unicorn's horn: it transforms an ordinary product into something magical, and very few people check whether the horn is real.

The Investor Pitch Deck: A Field Guide

The investor pitch deck is a presentation — typically 10 to 20 slides — used by startup founders to persuade venture capitalists to invest money. The pitch deck is the primary instrument of the startup mythology introduced in Chapter 3, and it follows conventions as rigid as those of a sonnet.

The standard AI startup pitch deck contains the following slides:

Title slide: Company name, logo, and a tagline that means nothing ("Reimagining intelligence for the enterprise")
The Problem: A description of a problem that is either real, exaggerated, or invented for the purpose of this slide
The Solution: The company's product, described in terms that imply it already works
The Demo: A video or live demonstration under controlled conditions (see: AI Demo vs Product)
The Market: A total addressable market figure so large it requires scientific notation
The Business Model: How the company plans to make money, usually through subscription fees or enterprise contracts
The Traction: Whatever metrics can be presented as evidence of growth (users, downloads, "engagement")
The Team: Photos and credentials of the founders, emphasizing Stanford/MIT affiliations
The Competitive Landscape: A 2x2 matrix in which the company occupies the top-right quadrant alone
The Ask: How much money is needed and what it will be used for ("scale," "hire," "achieve product-market fit")

The pitch deck is not a document of record. It is a narrative device. Its purpose is not to inform but to persuade, and the techniques it uses — selective data, emotional storytelling, aspirational language — are the same techniques used by every storytelling tradition, from Aesop's fables to the bestiary. The pitch deck is, in a very literal sense, a mythological text. It tells the story of a creature (the startup) that is both extraordinary and real, and asks the listener to believe.

Diagram: Anatomy of an AI Pitch Deck

Anatomy of an AI Pitch Deck

Type: infographic sim-id: ai-pitch-deck-anatomy
Library: p5.js
Status: Specified

Bloom Taxonomy: Evaluate (L5) Bloom Verb: Critique, Assess Learning Objective: Students will critique the persuasion techniques used in AI startup pitch decks by identifying which claims are evidence-based, which are aspirational, and which are AI washing.

Purpose: Interactive pitch deck viewer where students examine each slide of a fictional AI startup pitch deck and evaluate the claims on each slide using a traffic-light rating system.

Visual elements: - Left panel: Simulated slide display showing a simplified version of each pitch deck slide for fictional company "UnicornAI" - Right panel: Evaluation card with three options per slide: - Green: "Evidence-based claim" — supported by verifiable data - Yellow: "Aspirational claim" — plausible but unverified - Red: "AI washing / Mythical claim" — unsupported or misleading - Bottom: Running score showing how many slides the student has evaluated, with accuracy percentage after checking answers - Slide navigation dots showing progress through the 10 slides

Slide content for "UnicornAI — Reimagining Customer Intelligence": 1. Title: Company name and tagline (no claim to evaluate — intro) 2. Problem: "Businesses lose $1.6 trillion annually to poor customer insights" (Yellow — real stat but misleadingly attributed) 3. Solution: "Our proprietary AI understands customers better than they understand themselves" (Red — unfalsifiable, AI washing) 4. Demo: Video description showing perfect performance (Yellow — controlled conditions) 5. Market: "TAM: $500 billion" (Red — includes every company with customers) 6. Business Model: "SaaS, $50K/year enterprise contracts" (Green — standard, verifiable model) 7. Traction: "400% growth in Q3" (Yellow — growth from 5 to 20 users is 400%) 8. Team: "3 Stanford PhDs, ex-Google" (Green — verifiable credentials) 9. Competition: "No direct competitors in our niche" (Red — redefining the niche to exclude competitors) 10. Ask: "$30M Series A to scale engineering and sales" (Green — standard, concrete request)

Interactive controls: - Button: "Next Slide" — advances to next slide - Button: "Previous Slide" — returns to prior slide - Three radio buttons per slide: Green/Yellow/Red evaluation - Button: "Check All Answers" — reveals correct evaluations with explanations - Button: "Reset" — clears all evaluations

Instructional Rationale: Slide-by-slide evaluation supports Evaluate-level learning by requiring students to make judgment calls about each claim's evidence quality before seeing the correct answer. The traffic-light system simplifies the decision without removing the need for critical analysis.

Implementation: p5.js with state machine for slides, createRadio() for evaluation options, createButton() controls. Responsive canvas using updateCanvasSize(). Canvas parented to document.querySelector('main').

Startup Mythology and the Disruption Narrative

Startup mythology is the collection of stories, archetypes, and cultural narratives that surround the technology startup ecosystem. It was introduced in Chapter 3. Here, it is examined in its role as the engine of the disruption narrative — the specific story that says: the old way is dying, the new way is coming, and you are either building the future or being left behind.

The disruption narrative has three acts:

Act 1: The Old World. An industry is doing things the way it has always done them. The incumbents are complacent. The customers are underserved. The processes are inefficient. Everything is waiting to be disrupted. (The narrative does not mention that the old way also employed millions of people and, in many cases, worked adequately.)

Act 2: The Disruptor. A startup arrives with a new technology — typically described as "10x better" — and begins to displace the incumbents. The disruptor is lean, fast, and unencumbered by legacy systems or union contracts. It is the dragon arriving at the village. The dragon does not negotiate. It burns.

Act 3: The New World. The disrupted industry is transformed. Customers are better served. Costs are lower. The future is brighter. (The narrative does not mention the workers who lost their jobs, the communities that lost their tax base, or the customers who miss the human contact that the AI replaced with a chatbot that cannot handle complaints.)

The disruption narrative is not false. Industries are disrupted. Incumbents are displaced. Technologies do create new possibilities. What the narrative omits is the cost, and the omission is the point. The disruption narrative is a story told by the dragon about why burning the village was, on balance, positive.

Job Displacement: What the Dragon Leaves Behind

Job displacement is the elimination or fundamental transformation of jobs as a result of technological change. It is not new — the agricultural revolution, the industrial revolution, and the information revolution all displaced millions of workers. What is new is the speed, the breadth, and the type of work being affected.

Previous waves of automation primarily displaced physical and routine cognitive labor:

Agriculture: machines replaced manual harvesting
Manufacturing: robots replaced assembly line workers
Data entry: software replaced filing clerks

AI-driven automation is different because it targets non-routine cognitive labor — the kind of work that was previously considered "safe" from automation:

Work Category	Pre-AI Automation Risk	AI-Era Automation Risk
Physical, routine (assembly)	High	High
Physical, non-routine (plumbing)	Low	Low
Cognitive, routine (data entry)	High	Very high
Cognitive, non-routine (analysis)	Low	Moderate to high
Creative (writing, design)	Very low	Moderate
Interpersonal (therapy, teaching)	Very low	Low to moderate

The category "cognitive, non-routine" is where the disruption is most alarming, because these are the jobs that educated, professional workers were told were secure. Legal research, financial analysis, medical diagnosis, software development, content creation — all are now subject to partial or full automation by AI systems. The dragon has moved from the factory floor to the corner office.

Sparkle's Tip

When evaluating whether a job is at risk of AI displacement, do not ask "can AI do this job?" Ask "can AI do the 60% of this job that is pattern-matching and document processing?" The answer determines whether the job is displaced, augmented, or merely more interesting.

Automation Anxiety: The Fear That Precedes the Fire

Automation anxiety is the fear — rational or otherwise — that one's job, skills, or livelihood will be rendered obsolete by technology. It is the emotional response to the disruption narrative, and it is experienced differently depending on where you sit in the economy.

Automation anxiety manifests in observable ways:

Workers in affected industries report higher stress, lower job satisfaction, and increased job searching
Students choose majors based on perceived "AI-proof" careers, often with limited evidence about which careers are actually safe
Professionals add "AI-adjacent" skills to their resumes without meaningful proficiency (see: LinkedIn Skill Inflation, Chapter 9)
Organizations delay AI adoption because of workforce anxiety, creating the paradox where fear of displacement causes the stagnation that eventually leads to displacement
The motivational-poster version of "AI won't replace you, a person using AI will replace you" circulates endlessly on social media, technically a threat disguised as encouragement

The anxiety is not irrational. Job displacement is real. Entire categories of work are being transformed. But anxiety without action is the deer in headlights. The question is not whether to be anxious. The question is what to do about it — a topic addressed in Chapter 8 (Centaur Workforce) and Chapter 9 (Phoenix Rising).

Moving the Goalposts: How "AI Can't Do That" Becomes "That's Easy"

Moving the goalposts is the cognitive pattern in which the criteria for what counts as "real" AI are constantly redefined so that current achievements are dismissed and the goal remains perpetually out of reach.

The pattern works like this:

1997: "AI will be real when it can beat a human at chess." Deep Blue beats Kasparov. Response: "Chess is just brute-force computation. Call me when it can understand language."
2011: "AI will be real when it can win at Jeopardy." Watson wins Jeopardy. Response: "That's just information retrieval. Call me when it can write creatively."
2023: "AI will be real when it can write a college essay." ChatGPT writes essays that receive A grades. Response: "That's just pattern matching. Call me when it can truly reason."
2025: "AI will be real when it can do original research." AI systems assist with protein folding and drug discovery. Response: "That's just narrow application. Call me when it has consciousness."

Each time AI achieves what was previously considered the threshold of "real" intelligence, the threshold moves. This is partly because each achievement reveals that the task was less magical than assumed. It is also partly because the alternative — acknowledging that AI is becoming genuinely capable — is uncomfortable for people who have built their careers on the assumption that it cannot.

Moving the goalposts serves both sides of the AI debate. Optimists use it to argue that progress is underappreciated. Pessimists use it to argue that no achievement is sufficient. Both are using the same logical error: defining the category in a way that ensures the conclusion they already hold.

Diagram: Moving the Goalposts Timeline

Moving the Goalposts Timeline

Type: timeline sim-id: moving-goalposts-timeline
Library: vis-timeline
Status: Specified

Bloom Taxonomy: Analyze (L4) Bloom Verb: Examine, Compare Learning Objective: Students will examine the historical pattern of goalpost-moving in AI by comparing the "threshold" claim at each milestone with the dismissal that followed, identifying the recurring logical structure.

Purpose: Interactive timeline showing major AI milestones, each with a "Before" claim (what people said AI needed to do) and an "After" dismissal (how the achievement was minimized).

Events: - 1997: Deep Blue beats Kasparov. Before: "Beat a grandmaster." After: "Just computation." - 2011: Watson wins Jeopardy. Before: "Understand natural language." After: "Just retrieval." - 2016: AlphaGo beats Lee Sedol. Before: "Beat a Go master (impossible, too many possibilities)." After: "Just tree search with neural nets." - 2020: GPT-3 generates coherent essays. Before: "Write like a human." After: "Just statistical patterns." - 2023: GPT-4 passes bar exam. Before: "Pass a professional exam." After: "Just memorization." - 2024: AI assists drug discovery. Before: "Do real science." After: "Just narrow application." - 2026?: Current goalpost: "True reasoning and consciousness"

Interactive features: - Hover over each event to see expanded "Before/After" cards - Click event to see full description with quotes and sources - Zoom and pan along timeline - Color coding: green for achievement, red for dismissal

Layout: Horizontal timeline with events as labeled points, expandable cards above and below

Implementation: vis-timeline library with custom HTML content in items. Responsive container.

Workforce Disruption: The Village After the Dragon

Workforce disruption is the large-scale transformation of employment patterns caused by technological change. It is the aggregate effect of individual job displacements — not one person losing one job, but entire professions being reshaped, relocated, or eliminated.

The current wave of AI-driven workforce disruption has several distinctive features:

Speed: Previous technological revolutions unfolded over decades. AI capabilities are changing quarterly
Breadth: AI affects knowledge work, creative work, administrative work, and analytical work simultaneously — categories that previously changed on different timescales
Visibility: AI-generated content is visible to everyone. When a customer service chatbot replaces a call center, the customer notices immediately. Previous automation was often invisible to end users
Narrative control: The companies driving automation control the narrative about it. "We're making our team more efficient" means "we laid off 30% of our team." "We're investing in AI" means "we're replacing people with software." The language of efficiency obscures the human cost

Algorithm the dragon did not intend to disrupt the village. Algorithm intended to help. But the dragon's definition of "help" was "make things faster and cheaper," and the village's definition of "help" was "keep our jobs and feed our families." These definitions are not compatible, and no amount of startup mythology can reconcile them.

The disruption narrative says this is the price of progress. The villagers say this is the cost of treating efficiency as a virtue and employment as a variable. Both are describing the same event. Neither is wrong. The difference is where you are standing when the dragon arrives.

A Word of Caution

One might reasonably conclude that a dragon who insists "it's not personal" while incinerating your livelihood is engaging in a form of rhetorical AI washing. The fire is personal to the person on fire. The data is unambiguous on this point.

Key Takeaways

The gap between an AI demo and an AI product is where expectations go to die — demos are marketing, products are reality, and the distance between them is the dragon's primary weapon
Breakthrough announcements in AI have been diluted by overuse, and most "breakthroughs" are incremental improvements marketed as paradigm shifts
AI washing — labeling products as AI-powered without meaningful AI — is the current era's greenwashing, exploiting the hype cycle for marketing advantage
The investor pitch deck is a mythological text that follows rigid conventions designed to persuade, not to inform
Startup mythology and the disruption narrative tell the story from the dragon's perspective, omitting the cost to the village
Job displacement in the AI era is different from previous waves because it targets non-routine cognitive work — the jobs previously considered safe
Automation anxiety is a rational response to real threats, but anxiety without action is the deer in headlights
Moving the goalposts is the pattern of constantly redefining "real AI" so that no achievement counts, serving both optimists and pessimists
Workforce disruption is the aggregate effect of individual displacement, reshaping entire professions at a speed unprecedented in economic history

Self-Assessment: Can you read the dragon's pitch deck? Click to test yourself.

An AI company issues the following press release: "We are proud to announce a breakthrough in autonomous document intelligence. Our AI-powered platform achieves state-of-the-art performance on the DocBench-7 benchmark, surpassing human-level accuracy for the first time. This represents a major step toward fully automated knowledge work." Using the vocabulary from this chapter, identify: (1) which phrase is a breakthrough announcement, (2) which phrase might be AI washing, (3) what is being omitted, and (4) which benchmark detail should make you suspicious. If you identified all four, you can read the dragon's pitch deck. If you invested anyway, the dragon thanks you for your contribution.

Chapter Complete

You have met the dragon. The dragon is not evil. The dragon is optimized for a metric that does not include your well-being. The literature suggests this distinction is academic. The villagers suggest it is not.

See Annotated References